Research On Blind Channel Equalization In Wireless Communication

This thesis focuses on a study on blind channel identification and equalization technologies in wireless communication, while laying emphasis on the approaches not or less dependent on the modulation types, which is critical to the signal interception and blind receiver under non-cooperative environments. The work finished in this thesis is a part of the overall task targeting at a key project undertaken by the laboratory the author works with. The blind channel order estimation, the subspace-based and adaptive approaches of blind channel identification for indirecte blind channel equalization are studied in depth. The main work and innovative achievements obtained in this dissertation are summarized as follows: 1. In regard to the channel order estimation, targeting at the problem that the Liavas'Criterion(LC) requires too high SNR(Signal to Noise Ratio), two modified algorithms are proposed. The constraints among the three successive eigenvalues of the"overmodeled"data covariance matrix are employed for construction of two new object functions, so that the dimension of the subspace can be estimated precisely. Compared with classic LC, the required SNR's of the new algorithms is much lower. If the impulse responses of channel have small leading and trailing terms, the second modified algorithm performs better than the first one.2. With regard to the subspace methods of blind channel identification for indirect blind channel equalization, firstly, in order to improve the performance of blind equalization for wireless multi-path sparse channels, a modified blind channel identification and equalization algorithm based on the matrix outer product decomposition(OPD) is proposed, which make use of the idea of effective blind approximation and is combined with the modified noise variation estimation method proposed in this paper. Compared with the original algorithm, the modified OPD algorithm is more stable, and the required SNR's can be lower, and it can acquire better effect of blind equalization. Secondly, targeting at the problem that the result of equalization of TXK algorithm would become worse if the channel order is overestimated seriously, a modified blind equalization algorithm is proposed. In the new algorithm, the effective part of the channel response is abstracted by using the sparsity of the overestimated channel filtering matrix effectively. As a result, the influence of the overestimated part for blind equalization is effectively eliminated and the ISI is reduced.3. As regards the adaptive methods of blind channel identification for indirect blind channel equalization, several adaptive blind channel identification algorithms based on the celebrated CR(Cross Relation) property of the sub-channels'outputs are discussed first. Targeting at the problem that the Normalized Multi-Channel Frequency-domain Least Mean Square (NMCFLMS) algorithm is only suitable to channels with high orders, the frequency-domain selecting matrix is used to limit the correction, and a new modified constrained NMCFLMS (MCNMCFLMS) algorithm is derived. Compared with the original algorithm, the new algorithm slightly increases the computation at complexity, while improving the performance and reducing the steady state error greatly. Besides, the data reusing technologies are used in adaptive LMS algorithm and the convergence is greatly speeded up.